package de.jetwick.snacktory;
import java.util.ArrayList;
import java.util.Arrays;
import java.util.Collection;
import java.util.Collections;
import java.util.Comparator;
import java.util.LinkedHashMap;
import java.util.LinkedHashSet;
import java.util.List;
import java.util.Map;
import java.util.Set;
import java.util.regex.Pattern;
import org.jsoup.Jsoup;
import org.jsoup.nodes.Document;
import org.jsoup.nodes.Element;
import org.jsoup.select.Elements;
import org.slf4j.Logger;
import org.slf4j.LoggerFactory;
/**
* This class is thread safe.
*
* @author Alex P (ifesdjeen from jreadability)
* @author Peter Karich
*/
public class ArticleTextExtractor {
private static final Logger logger = LoggerFactory.getLogger(ArticleTextExtractor.class);
// Interessting nodes
private static final Pattern NODES = Pattern.compile("p|div|td|h1|h2|article|section");
// Unlikely candidates
private String unlikelyStr;
private Pattern UNLIKELY;
// Most likely positive candidates
private String positiveStr;
private Pattern POSITIVE;
// Most likely negative candidates
private String negativeStr;
private Pattern NEGATIVE;
private static final Pattern NEGATIVE_STYLE =
Pattern.compile("hidden|display: ?none|font-size: ?small");
private static final Set<String> IGNORED_TITLE_PARTS = new LinkedHashSet<String>() {
{
add("hacker news");
add("facebook");
}
};
private static final OutputFormatter DEFAULT_FORMATTER = new OutputFormatter();
private OutputFormatter formatter = DEFAULT_FORMATTER;
public ArticleTextExtractor() {
setUnlikely("com(bx|ment|munity)|dis(qus|cuss)|e(xtra|[-]?mail)|foot|"
+ "header|menu|re(mark|ply)|rss|sh(are|outbox)|sponsor"
+ "a(d|ll|gegate|rchive|ttachment)|(pag(er|ination))|popup|print|"
+ "login|si(debar|gn|ngle)");
setPositive("(^(body|content|h?entry|main|page|post|text|blog|story|haupt))"
+ "|arti(cle|kel)|instapaper_body");
setNegative("nav($|igation)|user|com(ment|bx)|(^com-)|contact|"
+ "foot|masthead|(me(dia|ta))|outbrain|promo|related|scroll|(sho(utbox|pping))|"
+ "sidebar|sponsor|tags|tool|widget|player|disclaimer|toc|infobox|vcard");
}
public ArticleTextExtractor setUnlikely(String unlikelyStr) {
this.unlikelyStr = unlikelyStr;
UNLIKELY = Pattern.compile(unlikelyStr);
return this;
}
public ArticleTextExtractor addUnlikely(String unlikelyMatches) {
return setUnlikely(unlikelyStr + "|" + unlikelyMatches);
}
public ArticleTextExtractor setPositive(String positiveStr) {
this.positiveStr = positiveStr;
POSITIVE = Pattern.compile(positiveStr);
return this;
}
public ArticleTextExtractor addPositive(String pos) {
return setPositive(positiveStr + "|" + pos);
}
public ArticleTextExtractor setNegative(String negativeStr) {
this.negativeStr = negativeStr;
NEGATIVE = Pattern.compile(negativeStr);
return this;
}
public ArticleTextExtractor addNegative(String neg) {
setNegative(negativeStr + "|" + neg);
return this;
}
public void setOutputFormatter(OutputFormatter formatter) {
this.formatter = formatter;
}
/**
* @param html extracts article text from given html string. wasn't tested
* with improper HTML, although jSoup should be able to handle minor stuff.
* @returns extracted article, all HTML tags stripped
*/
public JResult extractContent(Document doc) throws Exception {
return extractContent(new JResult(), doc, formatter);
}
public JResult extractContent(Document doc, OutputFormatter formatter) throws Exception {
return extractContent(new JResult(), doc, formatter);
}
public JResult extractContent(String html) throws Exception {
return extractContent(new JResult(), html);
}
public JResult extractContent(JResult res, String html) throws Exception {
return extractContent(res, html, formatter);
}
public JResult extractContent(JResult res, String html, OutputFormatter formatter) throws Exception {
if (html.isEmpty())
throw new IllegalArgumentException("html string is empty!?");
// http://jsoup.org/cookbook/extracting-data/selector-syntax
return extractContent(res, Jsoup.parse(html), formatter);
}
public JResult extractContent(JResult res, Document doc, OutputFormatter formatter) throws Exception {
if (doc == null)
throw new NullPointerException("missing document");
res.setTitle(extractTitle(doc));
res.setDescription(extractDescription(doc));
res.setCanonicalUrl(extractCanonicalUrl(doc));
// now remove the clutter
prepareDocument(doc);
// init elements
Collection<Element> nodes = getNodes(doc);
int maxWeight = 0;
Element bestMatchElement = null;
for (Element entry : nodes) {
int currentWeight = getWeight(entry);
if (currentWeight > maxWeight) {
maxWeight = currentWeight;
bestMatchElement = entry;
if (maxWeight > 200)
break;
}
}
if (bestMatchElement != null) {
List<ImageResult> images = new ArrayList<ImageResult>();
Element imgEl = determineImageSource(bestMatchElement, images);
if (imgEl != null) {
res.setImageUrl(SHelper.replaceSpaces(imgEl.attr("src")));
// TODO remove parent container of image if it is contained in bestMatchElement
// to avoid image subtitles flooding in
res.setImages(images);
}
// clean before grabbing text
String text = formatter.getFormattedText(bestMatchElement);
text = removeTitleFromText(text, res.getTitle());
// this fails for short facebook post and probably tweets: text.length() > res.getDescription().length()
if (text.length() > res.getTitle().length()) {
res.setText(text);
// print("best element:", bestMatchElement);
}
res.setTextList(formatter.getTextList(bestMatchElement));
}
if (res.getImageUrl().isEmpty()) {
res.setImageUrl(extractImageUrl(doc));
}
res.setRssUrl(extractRssUrl(doc));
res.setVideoUrl(extractVideoUrl(doc));
res.setFaviconUrl(extractFaviconUrl(doc));
res.setKeywords(extractKeywords(doc));
return res;
}
protected String extractTitle(Document doc) {
String title = cleanTitle(doc.title());
if (title.isEmpty()) {
title = SHelper.innerTrim(doc.select("head title").text());
if (title.isEmpty()) {
title = SHelper.innerTrim(doc.select("head meta[name=title]").attr("content"));
if (title.isEmpty()) {
title = SHelper.innerTrim(doc.select("head meta[property=og:title]").attr("content"));
if (title.isEmpty()) {
title = SHelper.innerTrim(doc.select("head meta[name=twitter:title]").attr("content"));
}
}
}
}
return title;
}
protected String extractCanonicalUrl(Document doc) {
String url = SHelper.replaceSpaces(doc.select("head link[rel=canonical]").attr("href"));
if (url.isEmpty()) {
url = SHelper.replaceSpaces(doc.select("head meta[property=og:url]").attr("content"));
if (url.isEmpty()) {
url = SHelper.replaceSpaces(doc.select("head meta[name=twitter:url]").attr("content"));
}
}
return url;
}
protected String extractDescription(Document doc) {
String description = SHelper.innerTrim(doc.select("head meta[name=description]").attr("content"));
if (description.isEmpty()) {
description = SHelper.innerTrim(doc.select("head meta[property=og:description]").attr("content"));
if (description.isEmpty()) {
description = SHelper.innerTrim(doc.select("head meta[name=twitter:description]").attr("content"));
}
}
return description;
}
protected Collection<String> extractKeywords(Document doc) {
String content = SHelper.innerTrim(doc.select("head meta[name=keywords]").attr("content"));
if (content != null) {
if (content.startsWith("[") && content.endsWith("]"))
content = content.substring(1, content.length() - 1);
String[] split = content.split("\\s*,\\s*");
if (split.length > 1 || (split.length > 0 && !"".equals(split[0])))
return Arrays.asList(split);
}
return Collections.emptyList();
}
/**
* Tries to extract an image url from metadata if determineImageSource
* failed
*
* @return image url or empty str
*/
protected String extractImageUrl(Document doc) {
// use open graph tag to get image
String imageUrl = SHelper.replaceSpaces(doc.select("head meta[property=og:image]").attr("content"));
if (imageUrl.isEmpty()) {
imageUrl = SHelper.replaceSpaces(doc.select("head meta[name=twitter:image]").attr("content"));
if (imageUrl.isEmpty()) {
// prefer link over thumbnail-meta if empty
imageUrl = SHelper.replaceSpaces(doc.select("link[rel=image_src]").attr("href"));
if (imageUrl.isEmpty()) {
imageUrl = SHelper.replaceSpaces(doc.select("head meta[name=thumbnail]").attr("content"));
}
}
}
return imageUrl;
}
protected String extractRssUrl(Document doc) {
return SHelper.replaceSpaces(doc.select("link[rel=alternate]").select("link[type=application/rss+xml]").attr("href"));
}
protected String extractVideoUrl(Document doc) {
return SHelper.replaceSpaces(doc.select("head meta[property=og:video]").attr("content"));
}
protected String extractFaviconUrl(Document doc) {
String faviconUrl = SHelper.replaceSpaces(doc.select("head link[rel=icon]").attr("href"));
if (faviconUrl.isEmpty()) {
faviconUrl = SHelper.replaceSpaces(doc.select("head link[rel^=shortcut],link[rel$=icon]").attr("href"));
}
return faviconUrl;
}
/**
* Weights current element. By matching it with positive candidates and
* weighting child nodes. Since it's impossible to predict which exactly
* names, ids or class names will be used in HTML, major role is played by
* child nodes
*
* @param e Element to weight, along with child nodes
*/
protected int getWeight(Element e) {
int weight = calcWeight(e);
weight += (int) Math.round(e.ownText().length() / 100.0 * 10);
weight += weightChildNodes(e);
return weight;
}
/**
* Weights a child nodes of given Element. During tests some difficulties
* were met. For instanance, not every single document has nested paragraph
* tags inside of the major article tag. Sometimes people are adding one
* more nesting level. So, we're adding 4 points for every 100 symbols
* contained in tag nested inside of the current weighted element, but only
* 3 points for every element that's nested 2 levels deep. This way we give
* more chances to extract the element that has less nested levels,
* increasing probability of the correct extraction.
*
* @param rootEl Element, who's child nodes will be weighted
*/
protected int weightChildNodes(Element rootEl) {
int weight = 0;
Element caption = null;
List<Element> pEls = new ArrayList<Element>(5);
for (Element child : rootEl.children()) {
String ownText = child.ownText();
int ownTextLength = ownText.length();
if (ownTextLength < 20)
continue;
if (ownTextLength > 200)
weight += Math.max(50, ownTextLength / 10);
if (child.tagName().equals("h1") || child.tagName().equals("h2")) {
weight += 30;
} else if (child.tagName().equals("div") || child.tagName().equals("p")) {
weight += calcWeightForChild(child, ownText);
if (child.tagName().equals("p") && ownTextLength > 50)
pEls.add(child);
if (child.className().toLowerCase().equals("caption"))
caption = child;
}
}
// use caption and image
if (caption != null)
weight += 30;
if (pEls.size() >= 2) {
for (Element subEl : rootEl.children()) {
if ("h1;h2;h3;h4;h5;h6".contains(subEl.tagName())) {
weight += 20;
// headerEls.add(subEl);
} else if ("table;li;td;th".contains(subEl.tagName())) {
addScore(subEl, -30);
}
if ("p".contains(subEl.tagName()))
addScore(subEl, 30);
}
}
return weight;
}
public void addScore(Element el, int score) {
int old = getScore(el);
setScore(el, score + old);
}
public int getScore(Element el) {
int old = 0;
try {
old = Integer.parseInt(el.attr("gravityScore"));
} catch (Exception ex) {
}
return old;
}
public void setScore(Element el, int score) {
el.attr("gravityScore", Integer.toString(score));
}
private int calcWeightForChild(Element child, String ownText) {
int c = SHelper.count(ownText, """);
c += SHelper.count(ownText, "<");
c += SHelper.count(ownText, ">");
c += SHelper.count(ownText, "px");
int val;
if (c > 5)
val = -30;
else
val = (int) Math.round(ownText.length() / 25.0);
addScore(child, val);
return val;
}
private int calcWeight(Element e) {
int weight = 0;
if (POSITIVE.matcher(e.className()).find())
weight += 35;
if (POSITIVE.matcher(e.id()).find())
weight += 40;
if (UNLIKELY.matcher(e.className()).find())
weight -= 20;
if (UNLIKELY.matcher(e.id()).find())
weight -= 20;
if (NEGATIVE.matcher(e.className()).find())
weight -= 50;
if (NEGATIVE.matcher(e.id()).find())
weight -= 50;
String style = e.attr("style");
if (style != null && !style.isEmpty() && NEGATIVE_STYLE.matcher(style).find())
weight -= 50;
return weight;
}
public Element determineImageSource(Element el, List<ImageResult> images) {
int maxWeight = 0;
Element maxNode = null;
Elements els = el.select("img");
if (els.isEmpty())
els = el.parent().select("img");
double score = 1;
for (Element e : els) {
String sourceUrl = e.attr("src");
if (sourceUrl.isEmpty() || isAdImage(sourceUrl))
continue;
int weight = 0;
int height = 0;
try {
height = Integer.parseInt(e.attr("height"));
if (height >= 50)
weight += 20;
else
weight -= 20;
} catch (Exception ex) {
}
int width = 0;
try {
width = Integer.parseInt(e.attr("width"));
if (width >= 50)
weight += 20;
else
weight -= 20;
} catch (Exception ex) {
}
String alt = e.attr("alt");
if (alt.length() > 35)
weight += 20;
String title = e.attr("title");
if (title.length() > 35)
weight += 20;
String rel = null;
boolean noFollow = false;
if (e.parent() != null) {
rel = e.parent().attr("rel");
if (rel != null && rel.contains("nofollow")) {
noFollow = rel.contains("nofollow");
weight -= 40;
}
}
weight = (int) (weight * score);
if (weight > maxWeight) {
maxWeight = weight;
maxNode = e;
score = score / 2;
}
ImageResult image = new ImageResult(sourceUrl, weight, title, height, width, alt, noFollow);
images.add(image);
}
Collections.sort(images, new ImageComparator());
return maxNode;
}
/**
* Prepares document. Currently only stipping unlikely candidates, since
* from time to time they're getting more score than good ones especially in
* cases when major text is short.
*
* @param doc document to prepare. Passed as reference, and changed inside
* of function
*/
protected void prepareDocument(Document doc) {
// stripUnlikelyCandidates(doc);
removeScriptsAndStyles(doc);
}
/**
* Removes unlikely candidates from HTML. Currently takes id and class name
* and matches them against list of patterns
*
* @param doc document to strip unlikely candidates from
*/
protected void stripUnlikelyCandidates(Document doc) {
for (Element child : doc.select("body").select("*")) {
String className = child.className().toLowerCase();
String id = child.id().toLowerCase();
if (NEGATIVE.matcher(className).find()
|| NEGATIVE.matcher(id).find()) {
// print("REMOVE:", child);
child.remove();
}
}
}
private Document removeScriptsAndStyles(Document doc) {
Elements scripts = doc.getElementsByTag("script");
for (Element item : scripts) {
item.remove();
}
Elements noscripts = doc.getElementsByTag("noscript");
for (Element item : noscripts) {
item.remove();
}
Elements styles = doc.getElementsByTag("style");
for (Element style : styles) {
style.remove();
}
return doc;
}
private void print(Element child) {
print("", child, "");
}
private void print(String add, Element child) {
print(add, child, "");
}
private void print(String add1, Element child, String add2) {
logger.info(add1 + " " + child.nodeName() + " id=" + child.id()
+ " class=" + child.className() + " text=" + child.text() + " " + add2);
}
private boolean isAdImage(String imageUrl) {
return SHelper.count(imageUrl, "ad") >= 2;
}
/**
* Match only exact matching as longestSubstring can be too fuzzy
*/
public String removeTitleFromText(String text, String title) {
// don't do this as its terrible to read
// int index1 = text.toLowerCase().indexOf(title.toLowerCase());
// if (index1 >= 0)
// text = text.substring(index1 + title.length());
// return text.trim();
return text;
}
/**
* based on a delimeter in the title take the longest piece or do some
* custom logic based on the site
*
* @param title
* @param delimeter
* @return
*/
private String doTitleSplits(String title, String delimeter) {
String largeText = "";
int largetTextLen = 0;
String[] titlePieces = title.split(delimeter);
// take the largest split
for (String p : titlePieces) {
if (p.length() > largetTextLen) {
largeText = p;
largetTextLen = p.length();
}
}
largeText = largeText.replace("»", " ");
largeText = largeText.replace("»", " ");
return largeText.trim();
}
/**
* @return a set of all important nodes
*/
public Collection<Element> getNodes(Document doc) {
Map<Element, Object> nodes = new LinkedHashMap<Element, Object>(64);
int score = 100;
for (Element el : doc.select("body").select("*")) {
if (NODES.matcher(el.tagName()).matches()) {
nodes.put(el, null);
setScore(el, score);
score = score / 2;
}
}
return nodes.keySet();
}
public String cleanTitle(String title) {
StringBuilder res = new StringBuilder();
// int index = title.lastIndexOf("|");
// if (index > 0 && title.length() / 2 < index)
// title = title.substring(0, index + 1);
int counter = 0;
String[] strs = title.split("\\|");
for (String part : strs) {
if (IGNORED_TITLE_PARTS.contains(part.toLowerCase().trim()))
continue;
if (counter == strs.length - 1 && res.length() > part.length())
continue;
if (counter > 0)
res.append("|");
res.append(part);
counter++;
}
return SHelper.innerTrim(res.toString());
}
/**
* Comparator for Image by weight
*
* @author Chris Alexander, chris@chris-alexander.co.uk
*
*/
public class ImageComparator implements Comparator<ImageResult> {
@Override
public int compare(ImageResult o1, ImageResult o2) {
// Returns the highest weight first
return o2.weight.compareTo(o1.weight);
}
}
}